Scan: Will there be an at least 8.0 magnitude earthquake in California before 2035?

date	predicted	market price	confidence
2026-08-30	10%	27%	72%
2026-08-23	10%	27%	70%
2026-08-16	5%	25%	72%

1. Planner Fable 5.1 · Fermi decomposition + tool plan

Tools chosen: kalshi_direct kalshi_related polymarket_related wikipedia article_search claude_news code_execution

Sub-questions (Fermi decomposition)

What is the current Kalshi YES price and how thinly traded is it, given the 10-year lockup distorts long-dated contract pricing?
What does the USGS UCERF3 model estimate as the 30-year probability of an M≥8.0 earthquake in California, and what does that imply for the ~10-year window to Dec 2035 under a Poisson assumption?
Has California ever had a recorded M≥8.0 earthquake (e.g., are 1857 Fort Tejon and 1906 San Francisco officially catalogued at 7.9 or 8.0+ by USGS), and what magnitude source/rounding does the market use?
Would a Cascadia subduction zone megathrust rupture (M8-9) with an epicenter off the Northern California coast near Cape Mendocino qualify as 'California or its territorial waters', and what is the ~10-year probability of a Cascadia M8+ event whose epicenter lands in California waters?
What are the prices on other strikes in the KXEARTHQUAKECALIFORNIA series (e.g., M7.0+, M7.5+ before 2035) and are they internally consistent with a Gutenberg-Richter magnitude distribution?
Are there any recent studies or seismic sequences (e.g., Cape Mendocino 2024 M7.0, San Andreas stress accumulation) that materially change near-term M8+ probability estimates?

Planner reasoning

This is a decade-long tail-risk question dominated by geophysical base rates rather than news flow. The key drivers are the UCERF3 30-year probability for M8+ on California faults (~7%), the fact that California has never recorded an instrumentally confirmed M8.0+ (1857 and 1906 were ~7.9), and whether a Cascadia subduction zone rupture with an epicenter off Cape Mendocino would count under the resolution wording. The Kalshi price for a 10-year contract also embeds time-value/capital-lockup effects, so market price may be inflated relative to true probability.

2. Research Parallel tool calls

tool	status	time	items	summary
kalshi_direct	OK	0.2s	1	## This Market's Kalshi Data Before 2035 (KXEARTHQUAKECALIFORNIA-35) - Current price (probability): 33.00% - 7-day price change: +5.00% - 30-day price change: -15.00% - Average daily volume: 135 contracts - Price range: 7.00% - 65.00% - Data points: 50 days
kalshi_related	OK	3.0s	2	2 related markets / summaries. series KXEARTHQUAKECALIFORNIA: 0 markets (skipped 3 no-signal) \| keyword 'earthquake California': ok \| keyword 'earthquake magnitude': ok
polymarket_related	OK	3.0s	0	Scanned 100 active Polymarket markets, kept 0 matches. keyword 'earthquake California': 0 markets \| keyword 'earthquake 8.0': 0 markets \| keyword 'Cascadia': 0 markets
wikipedia	OK	3.1s	5	Fetched 5 Wikipedia entries (0 missing pages).
article_search	OK	3.3s	27	Fetched 27 unique articles across 3 queries (lookback=730d). 'California magnitude 8 earthquake probability USGS UCERF3': 10 articles \| 'Cascadia subduction zone megathrust probability Northern California': 10 articles \| 'San Andreas fault big one forecast': 10 articles
claude_news	OK	22.5s	12	## Findings UCERF3 Statewide M8+ Probability: - The likelihood that California will experience a magnitude 8 or larger earthquake in the next 30 years increased from about 4.7% in UCERF2 to about 7% in UCERF3. (https://southern.scec.org/ucerf) - The chance of a big earthquake (magnitude 8 or
code_execution	OK	38.0s	0	## Key Findings: P(M≥8.0 California Earthquake before 2035) Poisson conversion of source probabilities (~10-year window, now→2035): - UCERF3 statewide M8+ (7% per 30 yr) → annualized rate λ≈0.00242/yr → 2.39% over 10 years - Cascadia full-margin M9 (10–15% per 50 yr), scaled by 25–5

3. Evidence Brief Sonnet 5 · 5789 chars

# Event Will there be an at least 8.0 magnitude earthquake in California (or its territorial waters) before Dec 31, 2035? # Outcomes to forecast - Yes - No # Kalshi market anchor **Current YES price: 33.00%** (KXEARTHQUAKECALIFORNIA-35). Volatile: +5% over 7 days, but -15% over 30 days; range has spanned 7%–65% over 50 data points. Thin/illiquid: avg daily volume only 135 contracts. This 10-year lockup contract likely carries a liquidity/duration premium and retail overpricing of tail risk relative to model-implied probabilities (see below). [kalshi_direct] # Sub-question answers 1. **Kalshi YES price / liquidity** — 33% current, thinly traded (135 contracts/day avg), high volatility (7%-65% range), suggesting price is not a tight consensus and may be distorted by low liquidity and long duration. [kalshi_direct] 2. **UCERF3 30-yr M≥8 probability** — UCERF3 estimates ~7% over 30 years statewide (up from 4.7% in UCERF2), with southern San Andreas most likely locus. Poisson-scaled to the ~9-10 year window to 2035 gives ≈2.3%-2.4%. [claude_news, code_execution, SCEC] 3. **Historical M≥8 record** — Neither 1857 Fort Tejon nor 1906 San Francisco officially reached M8.0; both are catalogued by USGS at M7.9. California has never had an *official* recorded M≥8.0 onshore/near-shore earthquake. [claude_news/USGS event pages] 4. **Cascadia qualification** — Cascadia subduction zone begins offshore near Cape Mendocino, Northern California, and can produce M8-9 events; some rupture epicenters could plausibly fall within California waters. USGS/UCERF2-era estimate: ~12% chance over 30 years of M≥8 Cascadia megathrust event attributable to Northern California; newer USGS Pacific NW report cites 10-15% chance of ~M9 Cascadia event in 50 years (whole margin, not CA-specific). Fraction of that hazard landing in CA waters is uncertain (analysts scale 25-50%). [claude_news, code_execution] 5. **Other Kalshi strikes** — KXEARTHQUAKECALIFORNIA-28 (Before 2028, M8+) prices at 10%, consistent with a shorter window implying lower cumulative probability than the 2035 contract's 33%. No M7.0+/7.5+ strikes found in the series. [kalshi_related] 6. **Recent seismic activity** — Dec 2024 Cape Mendocino M7.0 (with 120+ aftershocks) is the most notable recent large event; no material change to UCERF3 M8+ probability found. No 2024-2025 studies revising California-specific M8+ probability were located; most recent USGS update (Sept 2025) concerns Pacific Northwest/Cascadia broadly, not CA-specific revision. [article_search, claude_news] # Key facts (high-confidence, factual) 1. [kalshi_direct] Current YES = 33%, avg volume 135 contracts/day, 50-day range 7%-65%. 2. [USGS] Largest recorded California earthquakes (1857, 1906) are both M7.9 — no confirmed M8.0+ event in the modern instrumental/historical record. 3. [SCEC/UCERF3] Statewide 30-yr M8+ probability = ~7%; southern San Andreas most likely source. 4. [Wikipedia/USGS] Cascadia subduction zone capable of M8-9, extends offshore to Cape Mendocino, Northern California. 5. [kalshi_related] KXEARTHQUAKECALIFORNIA-28 (shorter horizon) priced at only 10%, versus 33% for the -35 contract. # Cross-market signals - Kalshi related: 2028 contract (same M8 threshold) = 10%; 2035 contract = 33% — a large jump not fully explained by extra ~7 years of hazard exposure alone, suggesting either liquidity premium, differing crowd risk perception, or genuine compounding (Cascadia/San Andreas overdue narratives). - Kalshi related: Japan M8+ before 2030 priced at 38% — higher than California, consistent with Japan's higher subduction-zone hazard. - Polymarket: No matching markets found (0 hits). - Sportsbook implied: N/A (not applicable to this event type). # Analyst opinions and speculation - Quantitative synthesis (code_execution) combining UCERF3 statewide M8+ (Poisson-scaled ~2.4%/10yr) with partial Cascadia contribution yields **combined P(YES) ≈ 4.6%-8.3%, central ~6%** for the 10-year window to 2035. - GR (Gutenberg-Richter) sanity check suggests UCERF3's 7%/30yr is conservative-to-reasonable, not an underestimate, given CA fault segmentation caps effective max magnitude below pure GR extrapolation. - Quaketrace/3rd-party: San Andreas fault "maxes out" near M8.3, implying nonzero but bounded tail risk from that source alone. # Directional lean per outcome - **Yes**: Supported by nonzero UCERF3 rate, Cascadia megathrust potential (M8-9, epicenter could count as CA waters), 10-year horizon extending exposure; opposed by no historical instrumented precedent of CA M8+, model-implied probability (~5-8%) far below current Kalshi price (33%). - **No**: Strongly supported by base-rate modeling (~92-95% implied), lack of any confirmed M8+ CA event in recorded history (250+ years), UCERF3's modest 7%/30yr figure; opposing risk is tail uncertainty around Cascadia epicenter attribution and long-tail modeling uncertainty. # Gaps / unknowns - No explicit resolution guidance on which magnitude catalog/version (preliminary vs. moment magnitude revisions) Kalshi will use — historical quakes are borderline (7.9) and future revisions could matter. - Ambiguity on whether a Cascadia rupture epicentered slightly offshore/north of the CA border but affecting CA counts as "California or territorial waters." - No Polymarket cross-check available. - No CA-specific USGS forecast update post-2015 (UCERF3) identified in search. # Calibration anchors - Kalshi current YES price (anchor): 33% - Model-based base rate estimate: ~5%-8% (central ~6%) for 10-year window - Precedent: KXEARTHQUAKECALIFORNIA-28 (shorter horizon, same threshold) priced at 10% - Historical frequency: 0 confirmed M8.0+ CA earthquakes in ~250 years of record (implies base rate is low but nonzero given plate tectonics/Cascadia)

4. Forecasters Fable 5.1 + GPT-5.6 Sol · in parallel

claude-fable-5-1

62% conf

Yes	11%
No	89%

UCERF3 gives ~7% probability of a statewide M8+ over 30 years, which Poisson-scales to roughly 2.4% for the ~10-year window to 2035; adding a partial Cascadia contribution (rupture epicentered in Northern California waters) yields a model-implied ~5-8%. California has never had a confirmed M8.0+ event in ~250 years of record (1857 and 1906 are both catalogued at M7.9). The Kalshi anchor of 33% is thinly traded (135 contracts/day), has swung 7%-65%, and reflects a 10-year lockup where the time value of capital and retail tail-risk overpricing systematically inflate YES on long-dated low-probability contracts; the shorter 2028 strike at 10% is also inconsistent with 33% for 2035. I shade above the pure model estimate to account for Cascadia attribution ambiguity, possible magnitude-catalog revisions (e.g., a preliminary M8.0 later revised down still counting), and general model tail uncertainty, landing at ~11% YES.

gpt-5.6-sol

73% conf

Yes	9%
No	91%

The 33% Kalshi YES anchor is weak because trading is thin, its range has been extremely volatile, and the shorter 2028 contract's 10% price implies an implausibly steep increase over seven additional years. UCERF3's approximately 7% probability over 30 years scales to only about 2.4% through 2035, strongly favoring No. Allowing for a qualifying Cascadia rupture raises the quantitative estimate to roughly 5-8%, while California's lack of any officially recorded M8.0+ earthquake and the absence of a recent hazard revision constrain the upside. I assign 9% to Yes, modestly above the model range because territorial-water ambiguity, catalog revisions, and tail-model uncertainty could affect resolution.

5. Devil's Advocate Sonnet 5 · red-team critique

Both forecasters converge tightly (9-11%) around the model-implied 5-8% central estimate, but this creates a striking anomaly: neither seriously grapples with why the actual Kalshi market (33%, albeit thin) and even the shorter-horizon 2028 contract (10%) both sit meaningfully above their estimates—if the market is this systematically wrong, that itself deserves more scrutiny (e.g., is there private information, structural mispricing across ALL these contracts, or a flaw in the UCERF3 Poisson-scaling logic itself?).

Both treat the 2028-vs-2035 pricing gap (10% vs 33%) as pure evidence of market irrationality/liquidity premium, but an alternative reading is that the market is pricing in higher-than-UCERF3 subjective probability of Cascadia/San Andreas triggering events specifically in the medium term (7-10yr) that models underweight—neither forecaster stress-tests whether UCERF3's Poisson-scaling assumption (independent, memoryless earthquake timing) is appropriate given "overdue fault" narratives explicitly flagged in the brief.

Neither forecast adequately addresses the resolution-ambiguity gap flagged in the brief: "No explicit resolution guidance on which magnitude catalog/version... will be used" and historical events already sit at borderline M7.9 — a preliminary announcement of M8.0 that is later downgraded could still count as YES if resolution occurs before revision, which is a distinct upside tail risk from the Cascadia-attribution issue they both cite.

Both anchor "central ~6%" from the code_execution synthesis without independently interrogating the Cascadia attribution fraction (analysts scale 25-50% of a 10-15%/50yr basin-wide probability)—given this fraction alone spans a 2x range, the "5-8%" combined estimate may understate uncertainty width, yet both forecasters compress to point estimates (9-11%) rather than reflecting a wider plausible range or fatter tail.

Neither forecast considers that thin liquidity and long duration cut both ways: a 10-year lockup contract could just as easily underprice tail risk (investors avoiding illiquid 10-yr commitments, suppressing YES) as overprice it due to "retail tail-risk overpricing" — the brief doesn't provide clear evidence for which bias dominates, yet both forecasters assert the overpricing story without acknowledging this ambiguity undermines their confidence in dismissing the 33% anchor.

Both forecasts effectively ignore the Japan comparison (38% for M8+ before 2030) as a sanity check—if broadly comparable subduction-zone-adjacent markets price similarly higher than pure model outputs, this cross-market pattern (rather than CA-specific mispricing) might suggest structural underestimation by UCERF3-style models across multiple markets, which neither forecaster explores.

6. Aggregator Fable 5.1 · final calibrated forecast

11% (68% confidence)

Yes: 11% No: 89%

Both forecasters drew the same inference from the sub-questions: UCERF3's ~7%/30-yr statewide M8+ rate Poisson-scales to ~2.4% through 2035, and even a generous Cascadia contribution (rupture epicentered in California waters) only lifts the model-implied figure to ~5-8%; California has zero officially catalogued M8.0+ events in ~250 years. The 33% Kalshi anchor deserves heavy discounting: a 10-year lockup where NO pays roughly a treasury-like ~4%/yr leaves no incentive to sell YES down, volume is ~135 contracts/day, the price has ranged 7%-65%, and the 2028 strike at 10% is internally inconsistent with 33% for 2035. The critique's strongest points are the resolution ambiguities (preliminary magnitude announcements at 8.0 later revised down, territorial-water attribution for a Cascadia rupture) and non-Poisson 'overdue' fault behavior, which justify shading above the model but not toward the market; the Japan comparison is not diagnostic since Japan's M8+ recurrence genuinely supports high prices. I settle at 11% YES, slightly above the forecast mean to reflect the wider Cascadia-attribution and catalog-revision uncertainty flagged by the critique.

Pipeline Timing

Total pipeline time: 147.6s

Per-tool research timings shown in the Research section above.